3. APPENDIX A: Notes on reporting service contacts for non-uniquely identifiable clients

3.1. Background

The Community Mental Health Care (CMHC) NMDS is a national data collection that is built from mental health service contact records reported by each State and Territory jurisdiction. The basic concept underpinning the collection is a simple one, involving the following components:

  • each service contact recorded for an identified client is reported to the national data pool;
  • each service contact is attributable to an individual client by virtue of an anonymous unique patient identifier attached to the service contact record.

This allows each service contact to be linked to an individual client to whom it ‘belongs’, and provides a basis for reporting on the number of clients in receipt of community mental health care and the services that they receive.

However, since its inception in 2000-01, the data reported by jurisdictions for the CMHC NMDS have included a variable number of service contacts for non-uniquely identifiable clients. For the purposes of this current data extract specification, these are defined in the following terms:

Service contacts for non-uniquely identifiable clients refer to those mental health service contacts for which a person identifier was not recorded.

Typically, when such contacts are reported to the national data set, the person identifier number is not reported but the contact record includes details on one or more person-level data elements (for example, date of birth, sex, country of birth etc).

For some jurisdictions such contacts comprise up to 20% of total service contact records. For a significant number of such records, person-level data elements are also reported. It is evident that these contacts are genuine service contacts in respect of an individual client rather than ‘non patient care’ activities of community-based mental health services. In this respect, they are fully compliant with the scope of the CMHC NMDS. The only problem is that they have no unique patient identifier attached to the contact record.

A variety of factors underlie the reporting of service contacts for non-uniquely identifiable clients. These include:

  • The service provider clinician may record a service contact for a person who has not been assigned a local patient identifier, and for whom there is no intention to assign such an identifier.
  • The clinician may record a service contact for a person prior to a local patient identifier being assigned to that individual.
  • Alternatively, a clinician may record a service contact for a person for whom a local patient identifier has been assigned but, due to technical limitations, the identifier is not included in the data submitted to the relevant State or Territory health department.

Differences between jurisdictions in how unregistered contact data are handled also contribute to the variation that is evident in the national data set. Three key differences that affect the national collection are:

  • Differences in patient registration guidelines—jurisdictions differ in the extent to which policy documentation has been developed to guide local agencies on when to assign patient identifiers in community settings. Some jurisdictions have prepared detailed instructions for agencies that specify the requirement for patient identifiers to be assigned to all clients for whom a clinically significant service is provided, regardless of whether the client is accepted for ongoing services. For others, decisions about when to assign identifiers are left to the discretion of local agencies.
  • Different approaches to whether service contacts for non-uniquely identifiable clients are held in State and Territory-level collections—while some jurisdictions require such contacts to be reported to the State/Territory health department, others do not, or have no capacity to store or process such data.
  • Different approaches by states and territories to whether service contacts for non-uniquely identifiable clients are included in data submitted for the CMHC NMDS—for those jurisdictions that hold such contact data within the State- or Territory-level collections, differences exist as to whether such records are included in the data submitted for National Minimum Data Set purposes.

3.2. How contacts for non-uniquely identifiable clients should be reported for CMHC extract purposes

Contacts for non-uniquely identifiable clients create an anomaly for the national analysis and reporting of data submitted by states and territories. The following approach to the reporting of non-uniquely identifiable client contacts has been incorporated in the extract specifications to promote consistency where these contacts are reported:

  • States and territories that choose to report service contacts for non-uniquely identifiable clients to the CMHC NMDS may continue to do so, following the requirements of the extract specification. There is no expectation that jurisdictions that have not previously submitted such data will do so in the future—the choice is entirely the responsibility of each State and Territory.
  • Where such records are submitted to the national collection, they will conform to the data model underpinning the extract—that is, for each unregistered client contact record, there must be a matching Person details record.
  • Service contacts for non-uniquely identifiable clients will be clearly differentiated through the addition of a Person identifier flag on all Service contacts and Person details records.
  • For each Service contact details record that represents a contact with a non-uniquely identifiable client, a unique patient identifier should be created to fill the Patient identifier field, which can be used in conjunction with the Person identifier flag to identify these contacts. All other data elements on the Service contact record should be reported where these are available.
  • For each unique patient identifier generated for a non-uniquely identifiable client reported in Service contact details records, a corresponding Person details record should be created. The various data elements normally included in Person details records should be reported here where these are available.

3.3. Options for how patient identifiers for non-uniquely identifiable clients can be generated

There are many ways to achieve this. Below are suggestions for two common tools—SPSS and Microsoft Access. Further advice in relation to other packages can be provided to jurisdictions should this be required.

Note that it does not matter if patient identifier values for non-uniquely identifiable clients clash with real identifiers since there is a flag that allows them to be differentiated.

The general method suggested below assumes that the data source for non-uniquely identifiable clients is contained in a file/table with a single row containing the demographic and contact details. It also assumes that you have already created the registered client data set and that these are still in a format manipulable by your software tool of choice.

The steps are:

3.3.1. SPSS

1. Add a unique identifier to each row if one does not exist. You can use the following idiom to create the values:

COMPUTE PersIdent =$CASENUM.

2. Add the data from the unregistered clients to the tables/files/spreadsheets containing the registered client data making sure that the Person identifier flag is set correctly:

SAVE OUTFILE <temporary path for unregistered data>.

ADD FILES=\* /FILE=<path to your registered **person** details data>
/KEEP <space separated list of **person** details items>.

SAVE OUTFILE <path to your final **person** details data>.

GET FILE <temporary path for unregistered data>.

ADD FILES=\* /FILE=<path to your registered **service contact**
details data> /KEEP <space separated list of **service contact**
details items>.

SAVE OUTFILE <path to your final **service contact** details data>.

3.3.2. Microsoft Access

  1. Add a unique identifier to each row if one does not exist:
  • Create a table which contains all of the required columns including ‘Person Identifier’. Ensure that the ‘Person Identifier’ column is of type AutoNumber.
  • Append the patient data to that table excluding ‘Person Identifier’ from the list of fields to append. Each row will be given a unique number.

2. Add the data from the unregistered clients to the tables/files/spreadsheets containing the registered client data making sure that the Person identifier flag is set correctly. Run two append queries:

  • One to append the person details records from your unregistered client table to your final person details table (already containing your registered person details records).
  • The other to append the service contact details records to the final service contact details table.